By Henry Gale
Oct 31 - (The Insurer) - After ChatGPT went viral in 2022, insurers scrambled to understand how they could capitalise on opportunities from artificial intelligence and manage the risks that it introduced for their customers. After months of exploration, we may be getting much closer to clarity on the latter.
Several insurers, including WR Berkley, Cincinnati Financial and Frederick Mutual, have been exploring introducing AI exclusions in commercial liability policies, some of which have been filed with U.S. state regulators.
Verisk’s Core Lines Services, which provides standardised policy language for insurers, announced in July that it had made a multistate filing to help insurers address emerging risks, with a new optional generative AI exclusion top of the list.
JANUARY 2026: A KEY MOMENT
A key moment in determining how AI risks will be treated will come on January 1, 2026, the date from which Verisk’s clients will be able to implement its optional GenAI exclusion.
“Verisk’s core lines business does not monitor when or how our customers implement the material we create,” Joseph Lam, vice president of general liability, core lines at Verisk, told The Insurer.
“Due to the quickly evolving nature of GenAI exposures, we have received strong interest from many of our customers to create underwriting tools to address this emerging risk. The decisions on if and when to implement our material is at the discretion of each customer and is generally subject to their underwriting guidelines and/or risk appetite.”
He added: “These endorsements are tools that allow carriers to have greater underwriting flexibility, including being able to write a risk they may otherwise have declined due to potential generative AI exposure.”
Brokers will be likely to watch carefully whether insurers use the endorsements in that kind of way, enabling more of their clients to access coverage, or instead look to apply them across the board, with the effect of reducing their clients’ coverage overall.
WHICH INSURERS MIGHT START EXCLUDING GENAI RISKS?
In many cases, insurers that have authorised Verisk to make filings on their behalf will not need to make additional filings themselves to begin using the endorsements, including the optional GenAI exclusion.
But some companies have made filings with state insurance departments, which The Insurer has analysed, to either adopt, delay the adoption of or choose not to adopt the changes.
Subsidiaries of nine insurance groups have filed to adopt them, either from January 1 or later dates in 2026, including international listed companies such as Arch and Everest, regional carriers and local mutuals.
That these companies have made such filings does not necessarily mean they will introduce Verisk’s GenAI exclusions, which are optional endorsements. The companies may have done so because they plan to introduce other endorsements that were grouped together with the GenAI exclusions in Verisk’s recent filings.
Equally, other carriers beyond these could introduce the GenAI exclusions without making filings of their own.
Of those that made such filings that The Insurer was able to contact, all either declined to comment on whether they would be adopting the GenAI exclusions (including Arch) or did not immediately respond (including Everest).
INTACT: WE DON’T APPLY BROAD AI EXCLUSIONS
Subsidiaries of two companies filed to delay adopting the changes indefinitely, while two made filings that said they would not be adopting them. Among the latter was a subsidiary of Intact Insurance, which told The Insurer its approach to AI risks did not involve applying “broad exclusions”.
“As a specialty insurer, Intact Insurance closely follows emerging trends such as artificial intelligence and generative AI and their implications for the insurance industry,” Intact told The Insurer. “We take a specialised approach to tailor coverage to the unique needs of our customers rather than applying broad exclusions.”
It added: “We believe in engaging directly with our brokers and customers at every step of the process to understand their specific exposures and coverage needs. If a customer has potential AI-related risks, we initiate the conversation with our experts to assess those exposures and determine the most appropriate and tailored coverage solutions.”
UNDERWRITERS OF AI INSURANCE WELCOME EXCLUSIONS
Meanwhile, others have seen AI risks as a new opportunity for the insurance sector, with potential to be underwritten in a class of their own – not unlike the emergence of the cyber market more than 25 years ago.
Munich Re and startups Armilla and Testudo were developing standalone AI insurance products before Verisk introduced the new exclusions. They are now expecting to underwrite policies from January that fill the gaps in coverage introduced by the exclusions.
“This is a coming of age for the AI market,” Armilla founder and CEO Karthik Ramakrishnan told The Insurer, saying the exclusions were inevitable because commercial general liability policies were not designed to cover AI.
But the form the AI insurance market takes is still yet to be determined. Will insurers seek to introduce AI or GenAI exclusions widely in 2026, and will they be able to implement them without being pushed back by brokers?
Could another existing part of the insurance market, for example cyber, step in to fill any gaps introduced by new exclusions in commercial liability policies? Or is AI really set to be the new cyber, in a new risk class of its own that many companies will need coverage for?
This week, The Insurer has gone in depth on the AI and GenAI exclusions insurers have been exploring in recent months. Catch up on Monday’s introduction to the topic, Tuesday’s take on defining AI, Wednesday’s deep dive into wordings and Thursday’s perspectives from AI underwriters.
Comments